Igeekphone reported on May 7th that the blogger Digital Chat Station today released the detailed parameters of the Qualcomm Snapdragon 8 Elite 2, which is manufactured using TSMC’s N3p process.
This is TSMC’s third-generation 3nm process technology. Under the same power consumption, the N3P process can improve performance by 4%. At the same main frequency, its power consumption is reduced by 9%, and the overall transistor density is increased by 4%.
The Snapdragon 8 Elite 2CPU still adopts the self-developed Oryon CPU architecture, maintaining a design of 2 super-large cores and 6 large cores.
Integrated with the Adreno 840 GPU, the independent cache has been increased from 12MB to 16MB, and the computing power of the NPU has also been enhanced from 80TOPS to 100TOPS, significantly improving the AI performance.
This chip will also support the SME1/SVE2 instruction set. SME stands for “Scalable MultiMedia Extensions”, which is part of the Arm64 architecture. By providing more vector registers and a larger vector length, Enable the processor to handle multimedia and graphics-related computing tasks more efficiently.
SVE2 has added multiple instructions for computer vision, 5G, and multimedia acceleration, no longer limited to high-performance computing and machine learning scenarios. It can enhance the efficiency of the CPU when handling high-load AI and multimedia tasks, and the power consumption control will be more excellent.
As is customary, the Snapdragon 8 Elite 2 will make its debut in October this year, and the Xiaomi 16 series will be the first to adopt it.